Exploratory Agent

Exploratory Agent is an AI exploratory tester that lives in your browser's side panel. One click starts a short testing session on the page you're looking at. The agent tries realistic and edge-case inputs, disrupts flows (back, reload, double submit), checks the layout at phone width, and records each defect with steps to reproduce, expected and actual behaviour, and a screenshot. It also catches JavaScript exceptions, console errors and failing requests automatically. For QA teams: the agent maps the site, writes session charters you can review and edit, runs them, and triages everything into one report, with confirmed issues first and uncertain ones flagged for a human to verify. Summary or detailed view in the panel, a full HTML report you can print to PDF, CSV for Jira, Azure DevOps or Linear, Markdown, JSON, and one-click "Copy as issue". Safe by default. - Read-only mode blocks form submissions and other changes unless you turn it off. - The agent can't leave the site you're testing. - Logout links are blocked so your session survives. - Confirmation dialogs are dismissed automatically. - The agent works in its own tab, never yours. - You confirm you're allowed to test each site. Bring your own Anthropic API key. You see a cost estimate before every run and set a hard cap per session. The extension has no server: runs, reports and your key stay in your browser, and page content goes only to Anthropic's API. Export any run's configuration as YAML and run it with the open-source command-line version, for example in CI. Only test sites you own or have written permission to test, ideally a staging environment with test accounts.